Anti-money laundering platform for mining and analyzing data to identify money launderers

Abstract

A platform may obtain, for a first set of entities involved in a money laundering investigation, target entity information for a target entity and related entity information for a set of related entities. The platform may analyze the target entity information and the related entity information to identify money laundering candidates. The platform may determine one or more relationships indicating a degree of similarity between the target entity and the one or more related entities. The platform may generate a graph data structure that associates the target entity and the one or more related entities using the one or more relationships. The platform may determine a score for the target entity and one or more scores for the one or more related entities. The platform may provide a recommendation indicating whether the target entity and/or the one or more related entities are likely to be engaging in money laundering.

What is claimed is: 
     
       1. A method, comprising:
 obtaining, by a device and from a first data source, information associated with a first set of entities involved in a money laundering investigation,
 the information associated with the first set of entities including target entity information for a target entity and related entity information for a set of related entities; 
 
 obtaining, by the device and from a second data source, additional information associated with the target entity and/or the set of related entities; 
 determining, by the device, one or more relationships between the target entity and the set of related entities,
 the one or more relationships indicating a degree of similarity between the target entity and the set of related entities; 
 
 generating, by the device, a graph data structure that associates the target entity and the set of related entities using the one or more relationships; 
 obtaining, by the device, training information from a third data source,
 the training information including a collection of words and phrases that are associated with sentiment scores; 
 
 applying, by the device, one or more standardization techniques to store the training information in a unified format,
 the training information including one or more file types or file formats; 
 
 generating, by the device, a data model using the training information after applying the one or more standardization techniques; 
 validating, by the device, the data model using test information to verify that the data model correctly predicts that an entity engaged in money laundering,
 the test information being from a fourth data source and including information associated with an entity involved in money laundering; 
 
 retraining, by the device, the data model based on determining that the data model did not correctly predict that the entity engaged in money laundering; 
 processing, by the device, the retrained data model using data from the graph data structure as input into the retrained data model; 
 determining, by the device, a score for the target entity and/or a set of scores for the set of related entities using the retrained data model,
 the score and/or the set of scores to identify whether the target entity and/or the set of related entities are likely to be money launderers; and 
 
 generating and providing, by the device, a recommendation indicating whether the target entity and/or the set of related entities are likely to be engaging in money laundering based on determining the score and/or the set of scores. 
 
     
     
       2. The method of  claim 1 , where the target entity information, the related entity information, and/or the additional information includes information associated with a business and/or information associated with an employee of the business,
 the information associated with the business including at least one of:
 information indicating a business name, 
 information indicating a field in which the business operates, or 
 information indicating a geographic location in which the business operates, or 
 
 the information associated with the employee of the business including at least one of:
 information indicating an employee name, or 
 information indicating an employee job title. 
 
 
     
     
       3. The method of  claim 1 , where the second data source includes a news data source, a social media data source, and/or an internal data source of an entity associated with the money laundering investigation. 
     
     
       4. The method of  claim 1 , further comprising:
 analyzing, by the device, the information associated with the first set of entities to identify one or more related entities of the set of related entities as money laundering candidates. 
 
     
     
       5. The method of  claim 1 , further comprising:
 analyzing the additional information to identify one or more additional related entities and/or one or more additional relationships between the target entity and the one or more related entities; 
 updating the graph data structure to include information associated with the one or more additional related entities and/or the information associated with the one or more additional relationships; 
 determining one or more scores for the one or more additional related entities using the retrained data model; and 
 providing another recommendation indicating whether the one or more additional related entities are likely to be engaging in money laundering. 
 
     
     
       6. The method of  claim 1 , further comprising:
 determining a set of relevancy scores for a set of edges included in the graph data structure, 
 determining that one or more relevancy scores of the set of relevancy scores satisfy a relevancy threshold, and 
 removing one or more edges and/or one or more nodes associated with the one or more relevancy scores that satisfy the relevancy threshold. 
 
     
     
       7. The method of  claim 1 , further comprising:
 analyzing the training information using machine learning techniques. 
 
     
     
       8. A non-transitory computer-readable medium storing instructions, the instructions comprising:
 one or more instructions that, when executed by one or more processors, cause the one or more processors to:
 obtain, from a first data source, information associated with a first set of entities involved in a money laundering investigation,
 the information associated with the first set of entities including target entity information for a target entity and related entity information for a set of related entities; 
 
 obtain, from a second data source, additional information associated with the first set of entities involved in the money laundering investigation; 
 determine one or more relationships between the target entity and the set of related entities,
 the one or more relationships indicating a degree of similarity between the target entity and the set of related entities; 
 
 generate a graph data structure that associates the target entity and the set of related entities using the one or more relationships; 
 obtain training information from a third data source,
 the training information including a collection of words and phrases that are associated with sentiment scores; 
 
 apply one or more standardization techniques to store the training information in a unified format,
 the training information including one or more file types or file formats; 
 
 generate a data model using the training information after applying the one or more standardization techniques; 
 validate the data model using test information to verify that the data model correctly predicts that an entity engaged in money laundering,
 the test information being from a fourth data source and including information associated with an entity involved in money laundering; 
 
 retrain the data model based on determining that the data model did not correctly predict that the entity engaged in money laundering; 
 analyze a set of nodes included in the graph data structure to determine a set of scores based on the retrained data model; and 
 provide a recommendation indicating whether the target entity and/or the set of related entities are likely to be engaging in money laundering based on the set of scores. 
 
 
     
     
       9. The non-transitory computer-readable medium of  claim 8 , where the one or more instructions, that cause the one or more processors to determine the one or more relationships, cause the one or more processors to:
 compare the target entity information and the related entity information associated with a related entity of the set of related entities, 
 determine whether a social distance between the target entity and the related entity satisfies a threshold social distance, and 
 identify a relationship of the one or more relationships based on determining that the social distance between the target entity and the related entity satisfy the threshold social distance. 
 
     
     
       10. The non-transitory computer-readable medium of  claim 8 , where the one or more instructions, that cause the one or more processors to determine the one or more relationships, cause the one or more processors to:
 compare the target entity information and the related entity information associated with a related entity of the set of related entities, 
 determine the degree of similarity between the target entity and the related entity based on comparing the target entity information and the related entity information associated with the related entity, and 
 assign, to the related entity, a weight value indicating the degree of similarity between the related entity and the target entity. 
 
     
     
       11. The non-transitory computer-readable medium of  claim 8 , where the one or more instructions, when executed by the one or more processors, further cause the one or more processors to:
 compare the additional information and one or more nodes and edges of the graph data structure; 
 determine that the additional information satisfies a threshold level of similarity with the one or more nodes and edges of the graph data structure; and 
 verify the one or more nodes and edges of the graph data structure based on determining that the additional information satisfies the threshold level of similarity. 
 
     
     
       12. The non-transitory computer-readable medium of  claim 8 , where the one or more instructions, that cause the one or more processors to analyze the set of nodes included in the graph data structure, cause the one or more processors to:
 provide a node value included in the graph data structure as input for the retrained data model, 
 analyze the node value using the retrained data model, and 
 determine a score based on using the retrained data model to analyze the node value,
 the score to be used to determine whether the target entity and/or the set of related entities are likely to be engaging in money laundering. 
 
 
     
     
       13. The non-transitory computer-readable medium of  claim 8 , where the one or more instructions, that cause the one or more processors to obtain the training information from the third data source cause the one or more processors to:
 obtain, from the third data source, the training information, the training information being associated with a second set of entities,
 the third data source being a news data source or a social media data source, and 
 the training information including financial terms; 
 
 where the one or more instructions, that cause the one or more processors to generate the data model using the training information, cause the one or more processors to:
 generate the data model using the training information associated with the second set of entities, a deep learning analysis technique, and a bag of words technique; and 
 
 where the one or more instructions, that cause the one or more processors to analyze the set of nodes included in the graph data structure, cause the one or more processors to:
 determine the set of scores by using the set of nodes of the graph data structure as input for the retrained data model,
 the set of scores indicating whether node values in the graph data structure are associated with positive sentiment scores or negative sentiment scores. 
 
 
 
     
     
       14. The non-transitory computer-readable medium of  claim 8 , where the one or more instructions, that cause the one or more processors to analyze the set of nodes to determine the set of scores, cause the one or more processors to:
 analyze the set of nodes over a time interval to determine the set of scores,
 the set of scores indicating whether node values in the graph data structure are trending in a positive sentiment direction or in a negative sentiment direction during the time interval; and 
 
 where the one or more instructions, that cause the one or more processors to provide the recommendation, cause the one or more processors to:
 provide the set of scores for graphical display on a user interface of a device. 
 
 
     
     
       15. A platform, comprising:
 a processor; and one or more computing resources to:
 obtain, from a first data source, information associated with a first set of entities involved in a money laundering investigation,
 the information including target entity information for a target entity and related entity information for a set of related entities; 
 
 analyze the target entity information and the related entity information to identify one or more related entities of the set of related entities as money laundering candidates; 
 determine one or more relationships between the target entity and the one or more related entities,
 the one or more relationships indicating a degree of similarity between the target entity and the one or more related entities; 
 
 generate a graph data structure that associates the target entity and the one or more related entities using the one or more relationships; 
 obtain training information from a second data source,
 the training information including a collection of words and phrases that are associated with sentiment scores; 
 
 apply one or more standardization techniques to store the training information in a unified format,
 the training information including one or more file types or file formats; 
 
 generate a data model using the training information after applying the one or more standardization techniques; 
 validate the data model using test information to verify that the data model correctly predicts that an entity engaged in money laundering,
 the test information being from a third data source and including information associated with an entity involved in money laundering; 
 
 retrain the data model based on determining that the data model did not correctly predict that the entity engaged in money laundering; 
 determine a score for the target entity and one or more scores for the one or more related entities based on the retrained data model; and 
 provide a recommendation indicating whether the target entity and/or the one or more related entities are likely to be engaging in money laundering. 
 
 
     
     
       16. The platform of  claim 15 , where the one or more computing resources are to:
 analyze additional information to identify information relevant to the money laundering investigation; and 
 update a set of nodes included in the graph data structure to store the information relevant to the money laundering investigation. 
 
     
     
       17. The platform of  claim 15 , where the one or more computing resources are to:
 analyze the target entity information and/or the related entity information for a related entity of the one or more related entities to determine a degree of similarity between the target entity and the related entity, 
 assign, to the related entity, a weight value indicating the degree of similarity between the related entity and the target entity. 
 
     
     
       18. The platform of  claim 15 , where the one or more computing resources are to:
 analyze the target entity information and the related entity information to determine semantic relationships between the target entity and the one or more related entities. 
 
     
     
       19. The platform of  claim 15 , where the one or more computing resources are to:
 determine the score for the target entity and/or the one or more scores for the one or more related entities using the retrained data model. 
 
     
     
       20. The platform of  claim 15 , where the one or more computing resources, when analyzing the training information to generate the data model, are to:
 analyze the training information using:
 artificial intelligence, 
 machine perception, or 
 computer vision.